# US 11,308,320 B2 · Multi-segment text search using machine learning model for text similarity > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 11,308,320 B2 - Document ID: US11308320B2 - Status: Granted - Kind: B2 - Application number: 16/718,081 - Filed: 2019-12-17 - Published: 2020-06-18 - Granted: 2022-04-19 - Earliest claimed: 2018-12-17 - Assignees: Cognition IP Technology Inc. - Applicants: Cognition IP Technology Inc. - Inventors: Bryant Lee; Andrew Tjang; Andrew Perry Chu; Uday Pulleti - Web page: https://patentgrep.com/patent/US11308320B2 ## CPC - G06V 30/414: Extracting the geometrical structure, e.g. layout tree; Block segmentation, e.g. bounding boxes for graphics or text - G06F 40/226: Validation - G06V 30/413: Classification of content, e.g. text, photographs or tables - G06V 30/418: Document matching, e.g. of document images - G06F 16/93: Document management systems - G06F 40/194: Calculation of difference between files - G06F 40/30: Semantic analysis - G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) ## Abstract Systems and methods may be provided for performing a search on an input text block. The input text block may be split into a plurality of input text segments. A text similarity algorithm may be used to find similar stored text segments to each of the plurality of input text segments. ## Claims ### Claim 1 A computer-implemented method for performing text search using machine learning, comprising: receiving a text input, the text input comprising a patent claim; splitting the patent claim into clauses; performing, by a machine learning model, text similarity matching between each clause and a plurality of stored text portions, the stored text portions associated with a plurality of reference documents, wherein the text similarity matching generates a plurality of similarity scores measuring the similarity between each clause and the plurality of stored text portions; and parsing a set of office action documents to extract a plurality of reference information statements and a plurality claim text segments and a plurality of citation statements, wherein each claim text segment corresponds to one of the plurality of citation statements and wherein each corresponding citation statements comprises a mapping to one or more lines of text in a reference document that corresponds to one of the extracted reference information statements; determining, for each claim text segment, a comparison statement based on comparing the claim text segment and the corresponding citation statement, wherein the comparison statement identifies the similarity or dissimilarity between the claim text segment and the corresponding citation statement. ### Claim 2 (depends on 1) The computer-implemented method of claim 1, wherein the clauses and plurality of stored text portions comprise one or more word embeddings. ### Claim 3 (depends on 1) The computer-implemented method of claim 1, wherein the clauses and plurality of stored text portions comprise a tensor encoding. ### Claim 4 (depends on 1) The computer-implemented method of claim 1, further comprising outputting a claim chart associating each clause to one or more stored text portions. ### Claim 5 (depends on 1) The computer-implemented method of claim 1, further comprising: generating synthetic office action text based on the clauses and one or more stored text portions, the synthetic office action text identifying the disclosure of each of the clauses by at least one of the one or more stored text portions; displaying the synthetic office action text. ### Claim 6 (depends on 1) The computer-implemented method of claim 1, further comprising: receiving feedback on the quality of the outputted one or more stored text portions; updating a text similarity machine learning model based on the feedback. ### Claim 7 (depends on 1) The computer-implemented method of claim 1, further comprising: performing, by a first machine learning model, classification of the text input into a class; performing, by a second machine learning model, document similarity matching between the text input and a plurality of reference documents to identify a subset of similar reference documents; selecting a third machine learning model based on the class; performing, by the third machine learning model, text similarity matching between each clause and a plurality of stored text portions, the stored text portions associated with the subset of similar reference documents. ### Claim 8 (depends on 1) The computer-implemented method of claim 1, wherein the machine learning model comprises a neural network. ### Claim 9 A system comprising one or more processors, and a non-transitory computer-readable medium including one or more sequences of instructions that, when executed by the one or more processors, cause the system to perform operations comprising: receiving a text input, the text input comprising a patent claim; splitting the patent claim into clauses; performing, by a machine learning model, text similarity matching between each clause and a plurality of stored text portions, the stored text portions associated with a plurality of reference documents, wherein the text similarity matching generates a plurality of similarity scores measuring the similarity between each clause and the plurality of stored text portions; and parsing a set of office action documents to extract a plurality of reference information statements and a plurality claim text segments and a plurality of citation statements, wherein each claim text segment corresponds to one of the plurality of citation statements and wherein each corresponding citation statements comprises a mapping to one or more lines of text in a reference document that corresponds to one of the extracted reference information statements; determining, for each claim text segment, a comparison statement based on comparing the claim text segment and the corresponding citation statement, wherein the comparison statement identifies the similarity or dissimilarity between the claim text segment and the corresponding citation statement. ### Claim 10 (depends on 9) The system of claim 9, wherein the clauses and plurality of stored text portions comprise one or more word embeddings. ### Claim 11 (depends on 9) The system of claim 9, wherein the clauses and plurality of stored text portions comprise a tensor encoding. ### Claim 12 (depends on 9) The system of claim 9, wherein the non-transitory computer-readable medium further comprises instructions for: outputting a claim chart associating each clause to one or more stored text portions. ### Claim 13 (depends on 9) The system of claim 9, wherein the non-transitory computer-readable medium further comprises instructions for: generating synthetic office action text based on the clauses and one or more stored text portions, the synthetic office action text identifying the disclosure of each of the clauses by at least one of the one or more stored text portions; displaying the synthetic office action text. ### Claim 14 (depends on 9) The system of claim 9, wherein the non-transitory computer-readable medium further comprises instructions for: receiving feedback on the quality of the outputted one or more stored text portions; updating a text similarity machine learning model based on the feedback. ### Claim 15 (depends on 9) The system of claim 9, wherein the non-transitory computer-readable medium further comprises instructions for: performing, by a first machine learning model, classification of the text input into a class; performing, by a second machine learning model, document similarity matching between the text input and a plurality of reference documents to identify a subset of similar reference documents; selecting a third machine learning model based on the class; performing, by the third machine learning model, text similarity matching between each clause and a plurality of stored text portions, the stored text portions associated with the subset of similar reference documents. ### Claim 16 (depends on 9) The system of claim 9, wherein the machine learning model comprises a neural network.